import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import plotly.express as px
import missingno as msno

import warnings
warnings.filterwarnings('ignore')

df = pd.read_csv('EcomShipping.csv')
df.head(5).T

	0	1	2	3	4
ID	1	2	3	4	5
Warehouse_block	D	F	A	B	C
Mode_of_Shipment	Flight	Flight	Flight	Flight	Flight
Customer_care_calls	4	4	2	3	2
Customer_rating	2	5	2	3	2
Cost_of_the_Product	177	216	183	176	184
Prior_purchases	3	2	4	4	3
Product_importance	low	low	low	medium	medium
Gender	F	M	M	M	F
Discount_offered	44	59	48	10	46
Weight_in_gms	1233	3088	3374	1177	2484
Reached.on.Time_Y.N	1	1	1	1	1

df['Reached.on.Time_Y.N'] = df['Reached.on.Time_Y.N'].astype(str)

# Categoric data
df.describe(include='O')

	Warehouse_block	Mode_of_Shipment	Product_importance	Gender	Reached.on.Time_Y.N
count	10999	10999	10999	10999	10999
unique	5	3	3	2	2
top	F	Ship	low	F	1
freq	3666	7462	5297	5545	6563

# print out the categoric columns and its category 
for c in df.select_dtypes(exclude='number').columns.tolist():
    print(c, sorted(df[c].unique()))

Warehouse_block ['A', 'B', 'C', 'D', 'F']
Mode_of_Shipment ['Flight', 'Road', 'Ship']
Product_importance ['high', 'low', 'medium']
Gender ['F', 'M']
Reached.on.Time_Y.N ['0', '1']

# Numeric data
df.describe()

	ID	Customer_care_calls	Customer_rating	Cost_of_the_Product	Prior_purchases	Discount_offered	Weight_in_gms
count	10999.00000	10999.000000	10999.000000	10999.000000	10999.000000	10999.000000	10999.000000
mean	5500.00000	4.054459	2.990545	210.196836	3.567597	13.373216	3634.016729
std	3175.28214	1.141490	1.413603	48.063272	1.522860	16.205527	1635.377251
min	1.00000	2.000000	1.000000	96.000000	2.000000	1.000000	1001.000000
25%	2750.50000	3.000000	2.000000	169.000000	3.000000	4.000000	1839.500000
50%	5500.00000	4.000000	3.000000	214.000000	3.000000	7.000000	4149.000000
75%	8249.50000	5.000000	4.000000	251.000000	4.000000	10.000000	5050.000000
max	10999.00000	7.000000	5.000000	310.000000	10.000000	65.000000	7846.000000

Data cleaning (check)

msno.bar(df, color = 'orange')
plt.title('Checking for Missing Values\n', fontsize = 40)
plt.show()

heatmap of the data for checking the correlation between the features and target column.

plt.figure(figsize = (18, 7))
sns.heatmap(df.select_dtypes(include='number').corr(), annot=True, fmt='0.2f', 
            annot_kws={'size': 15}, linewidth=2, linecolor='orange')
plt.show()

Checking value counts of columns

# create columns list to check
cols = ['Warehouse_block', 'Mode_of_Shipment', 'Customer_care_calls', 'Customer_rating',
        'Prior_purchases', 'Product_importance', 'Gender', 'Reached.on.Time_Y.N']

plt.figure(figsize = (16, 20))

# plotting the countplot of each categorical column.
for i, col in enumerate(cols):
    if i <= 8:
        ax = plt.subplot(4, 2, i+1)
        sns.countplot(x = col, data = df, ax = ax, palette='rocket')
        plt.title(f"\n{col} Value Counts\n", fontsize = 20)

plt.tight_layout()
plt.show()

Exploring relation of categorical columns

object_columns = df.select_dtypes(include = ['object'])
object_columns.sample(5)

	Warehouse_block	Mode_of_Shipment	Product_importance	Gender	Reached.on.Time_Y.N
181	F	Ship	low	M	1
9942	D	Ship	low	F	1
9407	F	Ship	medium	M	0
141	B	Flight	low	M	1
1722	D	Ship	medium	F	1

Warehouse column and categories proportion

warehouse = object_columns['Warehouse_block'].value_counts().reset_index()
warehouse.columns = ['warehouse', 'value_counts']
fig = px.pie(warehouse, names = 'warehouse', values = 'value_counts', 
             color_discrete_sequence = px.colors.sequential.matter_r, width = 650, height = 400,
             hole = 0.5)
fig.update_traces(textinfo = 'percent+label')

Reach on time count in warehouse block

# 1 : NOT on time and 0: on time
plt.figure(figsize = (17, 6))
sns.countplot(data = df, x='Warehouse_block', hue = df['Reached.on.Time_Y.N'],  
              palette='rocket')
plt.show()

Gender

gender = object_columns['Gender'].value_counts().reset_index()
gender.columns = ['Gender', 'value_counts']
fig = px.pie(gender, names = 'Gender', values = 'value_counts',  
             width = 650, height = 400, hole = 0.4, 
             color_discrete_sequence=px.colors.sequential.RdBu)
fig.update_traces(textinfo = 'percent+label')

# 1 : NOT on time and 0: on time
plt.figure(figsize = (17, 6))
sns.countplot(x='Gender', hue = 'Reached.on.Time_Y.N', data = df, palette='rocket')
plt.show()

Shipment method

mode = object_columns['Mode_of_Shipment'].value_counts().reset_index()
mode.columns = ['Mode_of_Shipment', 'value_counts']
fig = px.pie(mode, names = 'Mode_of_Shipment', values = 'value_counts', 
             color_discrete_sequence = px.colors.sequential.RdBu, 
             width = 650, height = 400, hole = 0.5)
fig.update_traces(textinfo = 'percent+label')

Reach on time vs Shipping method

plt.figure(figsize = (17, 6))
sns.countplot(x='Mode_of_Shipment', hue = 'Reached.on.Time_Y.N', 
              data = df, palette='rocket')
plt.show()

Product importance

product_imp = object_columns['Product_importance'].value_counts().reset_index()
product_imp.columns = ['Product_importance', 'value_counts']
fig = px.pie(product_imp, names = 'Product_importance', values = 'value_counts',
             color_discrete_sequence = px.colors.sequential.RdBu, 
             width = 650, height = 400, hole = 0.5)
fig.update_traces(textinfo = 'percent+label')

Product importance Shipment on time

# 1 : NOT on time and 0: on time
plt.figure(figsize = (17, 6))
sns.countplot(x='Product_importance', hue = 'Reached.on.Time_Y.N', 
              data = df, palette='rocket')
plt.show()

Relation of continuous columns with on time or not

integer_columns = df.select_dtypes(include = ['int64'])
integer_columns.head()

	ID	Customer_care_calls	Customer_rating	Cost_of_the_Product	Prior_purchases	Discount_offered	Weight_in_gms
0	1	4	2	177	3	44	1233
1	2	4	5	216	2	59	3088
2	3	2	2	183	4	48	3374
3	4	3	3	176	4	10	1177
4	5	2	2	184	3	46	2484

Customer care calls

customer_care = integer_columns['Customer_care_calls'].value_counts().reset_index()
customer_care.columns = ['Customer_care_calls', 'value_counts']
fig = px.pie(customer_care, names = 'Customer_care_calls', 
             values = 'value_counts', width = 650, height = 400,
             color_discrete_sequence = px.colors.sequential.RdBu, hole = 0.4)
fig.update_traces(textinfo = 'percent+label')

Customer call vs Shipment on time

# 1 : NOT on time and 0: on time
plt.figure(figsize = (17, 6))
sns.countplot(data = df, x='Customer_care_calls',   
              palette='rocket', hue = 'Reached.on.Time_Y.N')
plt.show()

Customers' rating

customer_ratings = integer_columns['Customer_rating'].value_counts().reset_index()
customer_ratings.columns = ['Customer_rating', 'value_counts']
fig = px.pie(customer_ratings, names = 'Customer_rating', values = 'value_counts',
             color_discrete_sequence = px.colors.sequential.RdBu, 
             width = 650, height = 400, hole = 0.4)
fig.update_traces(textinfo = 'percent+label')

Customers rating vs Shipment on time

# 1 : NOT on time and 0: on time
plt.figure(figsize = (17, 6))
sns.countplot(x='Customer_rating', hue = 'Reached.on.Time_Y.N', 
              data = df, palette='rocket')
plt.show()

Prior purchase

prior_purchases = integer_columns['Prior_purchases'].value_counts().reset_index()
prior_purchases.columns = ['Prior_purchases', 'value_counts']
fig = px.pie(prior_purchases, names = 'Prior_purchases', values = 'value_counts',
             color_discrete_sequence = px.colors.sequential.RdBu, 
             width = 650, height = 400, hole = 0.4)
fig.update_traces(textinfo = 'percent+label')

Prior purchase vs shipment on time

# 1 : NOT on time and 0: on time
plt.figure(figsize = (17, 6))
sns.countplot(x='Prior_purchases', hue = 'Reached.on.Time_Y.N', data = df, palette='rocket')
plt.show()

Reach on time vs not on time

# 1 : NOT on time and 0: on time
reached_on_time_y_n = df['Reached.on.Time_Y.N'].value_counts().reset_index()
reached_on_time_y_n.columns = ['Reached.on.Time_Y.N', 'value_counts']
fig = px.pie(reached_on_time_y_n, names = 'Reached.on.Time_Y.N', values = 'value_counts',
             color_discrete_sequence = px.colors.sequential.RdBu, 
             width = 650, height = 400, hole = 0.4)
fig.update_traces(textinfo = 'percent+label')

Cost of the product

plt.figure(figsize = (15, 7))
ax = sns.histplot(df['Cost_of_the_Product'], bins = 100, color = 'orange', kde=True)

plt.show()

relation between cost of the product and shipment on time

# 1 : NOT on time and 0: on time
px.box(data_frame = df, x = 'Reached.on.Time_Y.N', y = 'Cost_of_the_Product', 
       color = 'Reached.on.Time_Y.N' )